1. Programmable Logic Devices, Threshold Logic
Unit-4

2. Topics: Basic PLD’s- ROM, PROM, PLA and PLD
Realization of Switching Functions using PLD’s
Threshold Logic: Capabilities of Threshold gate
Synthesis of threshold functions
Multi gate Synthesis

3. Introduction
A programmable Logic Device is an integrated circuit with internal logic gates that are connected through electronic fuses.
Programming the device involves the blowing of fuses along the paths that must be disconnected so as to obtain a particular configuration.
The word programming here refers to a hardware procedure that specifies the internal configuration of the device.

4. Introduction
The gates in a PLD are divided into an AND array and an OR array that are connected together to provide an AND-OR sum of product implementation.
The initial state of a PLD has all the fuses intact.
Programming the device involves the blowing of internal fuses to achieve a desired logic function.

5. ROM vs. PLA/PAL Programmable Connections Programmable Connections
Fixed
Programmable
Programmable
Inputs
AND array
Outputs
Connections
OR array
(decoder)
(a) Programmable read-only memory (PROM)
Programmable
Programmable
Fixed
Inputs
Outputs
Connections
AND array
OR array
(b) Programmable array logic (PAL) device
Programmable
Programmable
Programmable
Programmable
Inputs
Outputs
Connections
AND array
Connections
OR array
(c) Programmable logic array (PLA) device

6. Read-Only Memory
ROM

7. ROM Decoder ORs Produces minterms Produce SOP’s A B C D S2 S1 S0 S3 11 2 3 4 5 6 7 8 9 10 12 13 14 15
4:16
dec
Enb A
‘B’C’D’
‘B’C’D
‘B’CD’
‘B’CD
‘BC’D’
‘BC’D
‘BCD’
‘ BCD
B’C’D’
B’C’D
B’CD’
B’CD
B C’D’
B C’D
B C D’
B C D F 1 3 2 A B C D

8. ROM ROM A decoder A set of programmable OR’s D7 D6 D5 D4 D3 D2 D1 D0X

9. General Logic Implementation
Given a 2kxn ROM, we can implement ANY combinational circuit with at most k inputs and at most n outputs.
k-to-2k decoder will generate all 2k possible minterms
Each of the OR gates must implement a m()
Each m() can be programmed

10. Find a ROM-based circuit implementation for:
Example
Find a ROM-based circuit implementation for:
f(a,b,c) = a’b’ + abc
g(a,b,c) = a’b’c’ + ab + bc
h(a,b,c) = a’b’ + c
Solution:
Express f(), g(), and h() in m() format (use truth tables)
Program the ROM based on the 3 m()’s

11. Example
There are 3 inputs and 3 outputs, thus we need a 8x3 ROM block.
f = m(0, 1, 7)
g = m(0, 3, 6, 7)
h = m(0, 1, 3, 5, 7)
3-to-8 decoder 1 2 3 4 5 6 7 a b c f g h

12. ROM as a Memory
Read Only Memories (ROM) or Programmable Read Only Memories (PROM) have:
N input lines,
M output lines, and
2N decoded minterms.
Can be viewed as a memory with the inputs as addresses of data.

13. (Memories) Volatile: Non-Volatile: Random Access Memory (RAM):
SRAM "static"
DRAM "dynamic"
Non-Volatile:
Read Only Memory (ROM):
Mask ROM "mask programmable"
EPROM "electrically programmable"
EEPROM “electrically erasable electrically programmable"
FLASH memory - similar to EEPROM with programmer integrated on chip

14. ROM as Memory
Example: For input (A2,A1,A0) = 011, output is (F0,F1,F2,F3 ) = 0010.
What are functions F3, F2 , F1 and F0 in terms of (A2, A1, A0)?
Address
8x4 ROM D0 D1 D2 D3 D4 D5 D6 D7 A2 A1 A0 F3 F2 F1 F0 X 1 2 3 4 5 6 7
A[2:0]
F[3:0] 3 4

15. Programmable Logic
PAL, PLA

16. Programmable Logic Array
PLAs
Programmable Logic Array
Pre-fabricated building block of many AND/OR gates (or NOR, NAND) "Personalized" by making/ breaking connections among the gates.
General purpose logic building blocks.

17. PLA

18. PLA

19. PLA
A 3×2 PLA with 4 product terms.

20. Design for PLA: Example
Implement the following functions using PLA
F0 = A + B' C'
F1 = A C' + A B
F2 = B' C' + A B
F3 = B' C + A
Input Side:
1 = asserted in term
0 = negated in term
- = does not participate
Personality Matrix
Outputs
Inputs
Product t
erm
Reuse of
erms A 1 - B C F 2 3
A B
B C
A C
Output Side:
1 = term connected to output
0 = no connection to output

21. Example: Continued F0 = A + B' C' F1 = A C' + A B F2 = B' C' + A B
Personality Matrix

22. Constants
Sometimes a PLA output must be programmed to be a constant 1 or a constant 0.
P1 is always 1 because its product line is connected to no inputs and is therefore always pulled HIGH;
this constant-1 term drives the O1 output.
No product term drives the O2 output, which is therefore always 0.
Another method of obtaining a constant-0 output is shown for O3.

23. PALs
Programmable Array Logic
a fixed OR array.

24. PAL inputs 1st output section 2nd output Only functions with section
at most four
products can be
implemented
3rd output
section
4th output
section

25. PAL W = ABC + CD X = ABC + ACD + ACD + BCD
Y = ACD + ACD + ABD

26. END of Unit-4 (Part-1)

27. Threshold Logicshold Logic Unit-4(Part-2) (STLD Autonomous)

28. <Threshold element>
Logic design of sw functions constructed of electronic gates different type of switching element : threshold element.
Threshold element can be considered as a generalization of the conventional gates.
A large class of sw fn. can be realized by a single threshold element through lots of combinations of weights and threshold.
Threshold function : A fn f(x1, x2, ••• , xn) is defined as a threshold fn iff there exists a set of weights {w1, w2, ••• , wn} and a threshold T such that
f(x1, x2, ••• , xn) = 1, iff
f(x1, x2, ••• , xn) = 0, iff
,where T and weights wi’s are real number, xi’s are binary inputs, and each weight wi is associated with a particular input var xi.
A straightforward approach to identify a threshold fn is to derive from the truth table, a set of 2n linear simultaneous inequalities and to solve them.
w1 ••• wi ••• wn T
x1 ••• xi ••• xn f
<Threshold element>

29. Capabilities and limitation (A two input NOR gate)
Eg) f(x1, x2, x3) = (0,1,3)
0  T w3  T w2 < T w2+w3  T w1 < T w1+w3 < T w1+w2 < T w1+w2+w3 < T
Inequality( ) f
x1 x2 x3
w3>w2
w3>w1
w3>w1+w3  w1<0
To find more effective methods, we will study the properties of a Threshold fn.
Capabilities and limitation (A two input NOR gate) -1 -½ x1 x2
x1 x2 f
<NOR gate>
Since a NOR gate is strongly logically complete, any combinational sw. fn can be realized by threshold elements alone. Because of huge possible combinations of weights and threshold, lots of sw. fn can be realized by only one threshold element  threshold fn

30. Is every sw. fn a threshold fn? No.
Problem 1: Given a sw. fn f(x1, x2, ••• , xn), determine whether or not it is a threshold fn and if it is, find weights and a threshold.
Problem 2: If the given sw. fn is not a threshold fn, how many threshold elements are required to realized the given sw. fn.
Properties: A fn f(x1, x2, ••• , xn) is positive in a var xi iff
f(x1, x2, ••• , xi-1, 1, xi+1, ••• , xn)  f(x1, x2, ••• , xi-1, 0, xi+1, ••• , xn)
A fn f(x1, x2, ••• , xn) is negative in a var xi iff
f(x1, x2, ••• , xi-1, 0, xi+1, ••• , xn) < f(x1, x2, ••• , xi-1, 1, xi+1, ••• , xn)
A fn which is positive or negative in every var is called unate.
positive unate : positive in every var negative unate : negative in every var
A completely specified fn is unate iff any minimal S. of P. representation of the fn uses either literal xi or literal xi’ but not both for 1  i  n.
Eg) f(x1, x2, x3) = (0,1,3) = x1’x2’+x1’x3’(unate), x1’x2+x1x2’(not unate) 1 10 11 01 00
x1x2 x3

31. Theorem: all threshold functions are unate.
Proof: let f(x1, x2, ••• , xn) be a non unate fn. Suppose it is a threshold fn then there should be some xi such that f is neither positive or negative in xi. Since f ins non unate, there exist some input combination, say (a1, a2, ••• , an) such that
f(a1, a2, ••• , ai-1, 0, ai+1, ••• , an) = 1
f(a1, a2, ••• , ai-1, 1, ai+1, ••• , an) = 0
and some input combination say (b1, b2, ••• , bn) such that
f(b1, b2, ••• , bi-1, 0, bi+1, ••• , bn) = 0
f(b1, b2, ••• , bi-1, 1, bi+1, ••• , bn) = 1
Wi<0
Wi>0
contradiction

32. Theorem: Not all unate function are a threshold fn.
Proof) We need to give one example, consider f(x1, x2, x3, x4) = (3,7,11,12,13,14,15) = x1x2+x3x4  unate fn
f(1,1,0,0)=1  w1+w2  T
f(1,0,1,0)=0  w1+w3 < T
f(0,0,1,1)=1  w3+w4  T
f(0,1,0,1)=0  w2+w4 < T
contradiction!! The given fn is not a threshold fn.
Consider that fn f(x1, x2, ••• , xn) is realized by V1= {w1, w2, ••• , wn; T} whose inputs are x1, x2, ••• , xn. If one of input, say xj, is complemented, the function can be realizable by a single threshold element having a weight-threshold vector.
V2= {w1, w2, •••, -wj, ••• , wn; T-wj} whose inputs are x1, x2, •••, -xj, ••• , xn. 01 00 10 11
x1x2 x3x4 1
w1+w2>w1+w3  w2>w3
w3+w4>w2+w4  w3>w2
w1 ••• wj ••• wn T
x1 ••• xj ••• xn f
w1 ••• -wj ••• wn
T-wj g 

33. We have to prove that f and g are identical fn.
Case 1: xj=0 (xj’=1)
left side (f) right side (g) 1
Case 1: xj=1 (xj’=0)
Therefore, f and g are identical.

34. If a function is realizable by a single threshold element (i. e
If a function is realizable by a single threshold element (i.e. threshold function) then by an appropriate selection of complemented and uncomplemented input variables, it is realizable by an threshold element whose weight have any desired sign distribution.
Let a threshold fn f(x1, x2, ••• , xn) be positive in var xi and the weight-threshold vector
V= {w1, w2, ••• , wn; T}
,since f is positive in var xi, there exists a set of values
a1, a2, ••• , ai-1, ai+1, ••• , an for inputs x1, x2, ••• , xi-1, xi+1, ••• , xn such that
f(a1, a2, ••• , ai-1, 1, ai+1, ••• , an) = 1 and f(a1, a2, ••• , ai-1, 0, ai+1, ••• , an) = 0
Hence,
w1a1+w2a2+ ••• + wi-1ai-1+wi+wi+1ai+1+ ••• +wnan  T and w1a1+w2a2+ ••• + wi-1ai wi+1ai+1+ ••• +wnan < T
wi>0
The weight wi associated with a threshold function which is positive in var xi, is positive.
The weights associated with a threshold function which is a positive in all its vars, are all positive.
If a function f(x1, x2, ••• , xn) is threshold fn with V1= {w1, w2, ••• , wn; T}, is the complement f ’(x1, x2, ••• , xn) a threshold fn? Then what is weigh-threshold vector, V2?

35. We assume that the weighted sum is not equal to T for any input combination
wixi>T  f= wixi<T  f=0 multiply by (-1) to each side.
 (-wi)xi<-T  f=1 (f ’=0) (-wi)xi>-T  f=0 (f ’=1)
V2= {-w1, -w2, ••• , -wn; -T}
Linear separability
If we use the n-cube representation for n-var threshold functions and regard the vertices as points in n-D space,
w1x1+w2x2+ ••• +wnxn = T
Correspond to an (n-1)-D hyperspace which cuts through the n-cube.
Now, if w1x1+w2x2+ ••• +wnxn  T then f=1, if w1x1+w2x2+ ••• +wnxn < T then f=0, the hyper plane separates the true vertices from the false vertices  Threshold function = linearly separable function.
If n=2, w1x1+w2x2=T x1 x2

36. If n=3, w1x1+w2x2 +w3x3=T  Plane equation
True vertices
Linear equation
False vertices
XOR function
0110 f
x1 x2
OR function
0111
(1,0)
(0,0)
(1,1)
(0,1)
separable
Not separable
 XOR isn’t threshold function
 We can separate between
true and false vertices

37. 1. Test the given fn for unateness.  convert to a positive unate fn
Our objective is to present a procedure which will determine whether a given sw fn is a threshold fn, and if it is, will provide the value of weights and threshold
1. Test the given fn for unateness  convert to a positive unate fn
2. Try to find only positive weights for a positive unate fn using minimal true vertices and maximal false vertices.
3. Convert to original inputs and adjust weights and threshold.
Ex) f(x1, x2, x3, x4) = (0,1,4,5,6,7,8,9,11,12,13,14,15) 01 00 10 11
x1x2 x3x4 1
f = x2+x3’+x1x4
We can say
positive unate in x1, x2, x4 negative unate in x3
Let g(y1, y2, y3, y4) = f(x1, x2, x3’, x4) then
g(y1, y2, y3, y4) = y2+y3+y1y4

38. (b1, b2, ••• , bn) < (a1, a2, ••• , an) 01 00 10 11
y1y2 y3y4 1
false vertices
true vertices
maximal false vertices
minimal true vertices
f  0 wixi < T
f  1 wixi  T
Minimal true vertices of a positive unate fn: any true vertex, say (a1, a2, ••• , an) such that there is no other true vertex, say (b1, b2, ••• , bn) such that
(b1, b2, ••• , bn) < (a1, a2, ••• , an)
Maximal false vertices of a positive unate fn: any false vertex, say (a1, a2, ••• , an) such that there is no other false vertex, say (b1, b2, ••• , bn) such that
(b1, b2, ••• , bn) > (a1, a2, ••• , an)

39. w4<w3 w4<w2 w4<w1+w4 w1<w3 w1<w2
Conti. example)
w4<w3 w4<w2 w4<w1+w4 w1<w3 w1<w2
If we choose, w3=4, w4=2, w1=2, and w2=4
Ex) f(x1, x2, x3, x4) = (0,1,3,4,5,6,7,12,13) 01 00 10 11
x1x2 x3x4 1
f = x1’x2+x1’x3’+x1’x4+x2x3’
positive unate in x2, x4 negative unate in x1, x3
Let g(y1, y2, y3, y4) = f(x1’, x2, x3’, x4) then
g(y1, y2, y3, y4) = y1y2+y1y3+y1y4+y2y3 3
3-4 f 
y1 y2 y3 y4
x1 x2 x3 x4

40. If we choose, w1=6, w2=5, w3=4, w4=2, and T=801 00 10 11
y1y2 y3y4 1
false vertices
true vertices
w1 w2+w4 w3+w4
w2+w3 w1+w4 w1+w3 w1+w2
w1<w2+w3 w2+w4<w2+w3 w2<w1 w2+w4<w1+w3 w4 <w1 w4<w w3 <w1 w3+w4<w1+w2
w1>w2,w3,w4
w4<w3
If we choose, w1=6, w2=5, w3=4, w4=2, and T=8 8 -2 f 
y1 y2 y3 y4
x1 x2 x3 x4